Luca Superiori

Content-aware scheduling for video streaming over HSDPA networks

In this article we propose the implementation of a content-aware cross-layer optimization for video streaming in HSDPA networks. The video packets are marked at the IP layer based on the impact on the quality of experience at the application layer. In the core network these packets are mapped onto multiple logical paths with different quality of service accordingly to their marking. The proposed approach complies with the 3GPP releases newer than R99. It has been tested with an HSDPA system level simulator and H-264 video traces. The results of simulations show that the proposed content aware scheduling delivers over 1 dB Y-PSNR improvement compared to the standard round robin.

Cross-Layer Detection of Visual Impairments in H.264/AVC Video Sequences streamed over UMTS Networks

Incorrectly received packets in low-rate video sequences result in the loss of considerably large picture areas that have to be concealed. The performance of error concealment decreases with the size of the interpolated picture area. The incorrectly received packets may still contain some correct information that can be exploited at the decoder. In this work we propose the utilization of information from the link layer of UMTS (universal mobile telecommunications System) at the application layer for better pre-localization of errors in the bitstream domain. Syntax check together with the detection of impairments in the pixel domain decide which part of the incorrectly received packets will be concealed. We evaluate the results using error traces from the live UMTS network and H.264/AVC (advanced video coding) encoded video stream. Apart from reduced complexity facilitated by the cross-layer approach, the proposed method gains 1.09 dBs of Y-PSNR compared to a slice rejection mechanism.

Detection of pan and zoom in soccer sequences based on H.264/AVC motion information

Unsupervised detection of pan and zoom in soccer sequences allows automatic classification of shots and match analysis. In this work we propose a pan and zoom (both in and out) detector specifically designed for low resolution soccer sequences. Our implementation is based on the analysis of the distribution of the motion vectors, already available in the encoded sequence, among a specific subset of reliable MBs, selected by means of inexpensive image preprocessing.

Cross-Layer Optimization of Video Services over HSDPA Networks

In the third generation networks a quality of service is guaranteed for key applications, such as video streaming. However, all the packets belonging to one application are handled in the same way, even though they have different impact on the perceived quality. In this article we present a standard compliant cross layer optimization for video services. The importance of a packet, signalized at application level, is used to filter packets into different logical channels with different qualities of service. Simulations performed using a HSDPA system level simulator show that our implementation increases the video quality of over 0.6 dB.

Encoding optimization of low resolution soccer video sequences

In this article we propose an optimization method for encoding low resolution soccer video sequences. The considered video codec is the state of the art H.264/AVC, as recommended by the 3GPP specifications. The proposed optimization aims at maximizing the user perceived quality, considering the characteristic features of soccer videos. In the typical wide angle shots, about one fourth of each frame contains the grandstands. Even if, from the observerpsilas point of view, they can be considered as side information, the performed analysis showed that a prominent fraction of the resulting bandwidth is allocated to them. Our approach consists of a segmentation of the picture intended to recognize three components: the field, the audience and the remaining objects: the ball, the players and the field lines. The resulting map will be then utilized while encoding to optimize the bitrate assigned to each macroblock group, targeting the best resulting subjective perceived quality.

Error concealment analysis for H.264/advanced video coding encoded video sequences

Error concealment methods have become very important in particular when transmitting video streams over error prone wireless links. Often a retransmission of corrupted sequences is not possible and thus the receiver has to make the best out of the received stream. The contributions of this article are the following: firstly, a performance comparison of various error concealment strategies (straight decoding, slice level concealment and macroblock level concealment) is presented based on the detection of errors, the exact location of which is unknown. Secondly, an analytical treatment of the slice level concealment, resulting in a precise mathematical model is provided. Finally, further improvements are proposed by subjective methods based on visual inspection and comparison of their performance by means of simulations.ZusammenfassungFehlerverschleierungsverfahren sind zurzeit von großem Interesse, insbesondere bei der Übertragung von Videos über fehlerbehaftete Funkstrecken. Meist ist eine Zweitübertragung einer fehlerbehafteten Sequenz nicht möglich, und der Empfänger muss das Beste aus den Empfängerdaten machen. In diesem Artikel präsentieren wir zunächst eine vergleichende Leistungsanalyse verschiedener Verschleierungsverfahren (direktes Dekodieren, Verschleierung auf Slice-Ebene und Verschleierung auf Makro-Ebene), basierend auf einer Fehlererkennung, dessen exakte Fehlerposition in der Sequenz allerdings unbekannt ist. Zweitens präsentieren wir eine analytische Behandlung der Verschleierung auf Slice-Ebene, die zu einem exakten mathematischen Modell führt. Schließlich schlagen wir weitere Verbesserungeen basierend auf subjektiven Verfahren der visuellen Inspektion vor und vergleichen ihre Leistungsfähigkeit anhand von Simulationen.

Robust Error Handling for Video Streaming over Mobile Networks

The quality of video streaming in an error-prone environment suffers from packet loss. Since the loss of a packet typically affects a rather large picture area, the performance of error concealment is limited, too. In order to avoid the huge overhead caused by using smaller packet sizes, in this paper we propose and analyze a scheme utilizing the residual redundancy of the encoded video stream. This scheme has two components -(i) syntax analysis implemented at the decoder, allowing exacter localization of errors within the packet and (ii) entropy code resynchronization mechanism based on the out-of-band signalized length indicators. We show that the proposed approach provides substantial improvement in PSNR for the same rate, compared to the standard packet size reduction.

Analysis of video streaming with SP and SI frames in UMTS mobile networks

In this paper we discuss possible benefits of transmitting SI frames as an error resilience tool in UMTS video streaming. SP and SI frames can be used to stop temporal error propagation, as an alternative to the regular insertion of I frames. Since their application relays on feedback information from the mobile equipment, different delay scenarios have been investigated. The performed transmission simulations show enhancement of the rate-distortion behaviour when using SP and SI frames with reduced feedback times. The error traces were measured at transport block level in a live UMTS mobile network.

Smart sorting of H.264/AVC encoded sequences for applications over UMTS networks

In this paper we propose a novel sorting mechanism suitable for H.264/AVC NALUs used in applications over UMTS networks. Each packet contains encoded information elements referred to a certain number of MBs. Following the standard, all the encoded information associated to a single MB is stored sequentially. In this work we propose to group the bits referring to a certain information element type even if associated to different MBs. The benefit of the proposed method will be discussed for different scenarios.

Optimization of audience encoding in low-resolution soccer video sequences

Soccer is one of the most streamed video contents over the European mobile networks. However, technical limitations, such as stringent bandwidth constraints and reduced size of the mobile terminal displays, make the delivery of soccer videos with satisfactory quality a challenging task. In this work we propose the optimization of the encoding process by minimizing the bandwidth associated to the audience while preserving acceptable quality for the viewer. The macroblocks belonging to the audience are identified by means of a segmentation algorithm. For these macroblocks, the camera movements will be compensated using a single motion vector. The optimized sequence can be decoded by a baseline profile H.264/AVC decoder.